Session: Substance Abuse
Room: Upson B17
Time: Tue 15:00-16:30

Presenter: Allen Goodman (Wayne State University. Economics)

Discussant: Mick Tilford (University of Arkansas for Medical Sciences)

Abstract

Researchers have long sought to characterize and identify statistical risk factors for the use of addictive substances. Many databases and analyses have concentrated on single addictions such as smoking, alcohol, or drugs, and analysts have implicitly assumed that contemporaneous addictions are unrelated to each other. This assumption is almost certainly incorrect. It is essential to examine co-occurring addictions assuming that they may reflect some underlying psychological or physical factors, with methods that recognize that none of the addictions necessarily cause the others.

In a full econometric treatment of addictive substances, one might model the ingestion of alcohol, cigarettes, and/or drugs as at least a two part model, in which the quantities ingested, total expenditures on individual commodities, or all of them together, constitute a second stage, conditional on the decision to drink, smoke, or take drugs. The potentially joint nature of these addictions makes the first modeling stage an important study in itself. How much people drink, smoke, or use drugs, depends on whether they use substances or not, and in what combinations.

Using the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) database, designed and conducted as the primary source for information and data on the U.S. population for: (1) alcohol and drug use; (2) alcohol and drug abuse and dependence; and (3) associated psychiatric and other medical comorbidities, Goodman (2009) presents a multinomial logit modeling strategy that addresses the joint determination of smoking, drinking, and drug use. His model then predicts absolute and marginal probabilities, and looks at gender and age related differences.

This study looks at the quantities demanded of cigarettes and ethanol (derived from beer, wine, and liquor) conditional on their single or multiple uses of addictive substances. Particular attention is directed to selection biases related to the usage of two or more addictive substances. These particular joint biases have not been modeled and estimated in previous work.

Preliminary work suggests that much of the “action” occurs in the decision to smoke or drink, rather than in the quantities smoked or drunk conditional on that decision. In other words, the primary impacts of incomes and prices may relate to the decision to start or stop smoking and drinking, rather than marginal adjustments in quantity by those who continue to smoke and drink.

Key Terms

Substance abuse, multiple addictions, cigarette demand, alcohol demand

Authors:

Allen Goodman (Wayne State University. Economics)